Room‐Temperature Laser Synthesis in Liquid of Oxide, Metal‐Oxide Core‐Shells, and Doped Oxide Nanoparticles

HAL Id: hal-03229952

https://hal.archives-ouvertes.fr/hal-03229952

Submitted on 27 May 2021

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Nanoparticles

Vincenzo Amendola, David Amans, Yoshie Ishikawa, Naoto Koshizaki, Salvatore Scirè, Giuseppe Compagnini, Sven Reichenberger, Stephan

Barcikowski

To cite this version:

Vincenzo Amendola, David Amans, Yoshie Ishikawa, Naoto Koshizaki, Salvatore Scirè, et al..

Room-Temperature Laser Synthesis in Liquid of Oxide, Metal-Oxide Core-Shells, and Doped Ox- ide Nanoparticles. Chemistry - A European Journal, Wiley-VCH Verlag, 2020, 26 (42), pp.9206-9242.

�10.1002/chem.202000686�. �hal-03229952�

& Laser Chemistry

Room-Temperature Laser Synthesis in Liquid of Oxide, Metal-Oxide Core-Shells, and Doped Oxide Nanoparticles

Vincenzo Amendola,*

David Amans,

Yoshie Ishikawa,

Naoto Koshizaki,

Salvatore ScirH,

Giuseppe Compagnini,

Sven Reichenberger,

and Stephan Barcikowski*

Abstract: Although oxide nanoparticles are ubiquitous in science and technology, a multitude of compositions, phases, structures, and doping levels exist, each one requir- ing a variety of conditions for their synthesis and modifica- tion. Besides, experimental procedures are frequently domi- nated by high temperatures or pressures and by chemical contaminants or waste. In recent years, laser synthesis of col- loids emerged as a versatile approach to access a library of clean oxide nanoparticles relying on only four main strat- egies running at room temperature and ambient pressure:

laser ablation in liquid, laser fragmentation in liquid, laser melting in liquid and laser defect-engineering in liquid. Here, established laser-based methodologies are reviewed through the presentation of a panorama of oxide nanoparticles

which include pure oxidic phases, as well as unconventional structures like defective or doped oxides, non-equilibrium compounds, metal-oxide core–shells and other anisotropic morphologies. So far, these materials showed several useful properties that are discussed with special emphasis on cata- lytic, biomedical and optical application. Yet, given the end- less number of mixed compounds accessible by the laser-as- sisted methodologies, there is still a lot of room to expand the library of nano-crystals and to refine the control over products as well as to improve the understanding of the whole process of nanoparticle formation. To that end, this review aims to identify the perspectives and unique oppor- tunities of laser-based synthesis and processing of colloids for future studies of oxide nanomaterial-oriented sciences.

Introduction

Oxide nanoparticles (NPs) are largely exploited for a variety of purposes, which embraces fields as different as, for instance, heterogeneous catalysis, biotechnology, medicine, photonics, solar energy conversion, microelectronics, automotive industry, pharmaceutics, and food additives.^[1,2] This variety of applica- tions also comes with a vast number of distinct compounds belonging to the class of oxide nanomaterials. The synthesis of

oxides with tailored properties requires a multitude of different synthetic procedures, including for instance the hydrothermal, calcination, mini-emulsion, spray pyrolysis, plasma-assisted and inert-atmosphere growth methods.^[1–3] Usually, these proce- dures allow high productivity but require sophisticated setups (autoclave reactors for the hydrothermal methods, furnaces for calcination, vacuum systems for plasma-assisted and inert at- mosphere methods, pressure- and precursor-flow-controlled flame synthesis). Also tailored experimental conditions includ- ing high temperature and pressure (hydrothermal, calcination, spray pyrolysis) and chemical precursors and/or additives that potentially persist as contaminants in the final products (calci- nation, mini-emulsions, plasma-assisted chemical vapor deposi- tion) are needed. The required precursors and their synthesis, as well as the post-treatment, often leads to toxic or pollutant waste, which poses the problem of their disposal.^[3]Moreover, for some oxide cations, no precursors are available at all, limit- ing their flame spray synthesis. In the framework of the global efforts towards a circular and sustainable economy, it is there- fore of utmost importance to develop synthesis routes running at room temperature and ambient pressure, which allows the cost-effective and green development of nanotechnologies based on oxides.

To this end, the laser-assisted synthesis of colloidal NPs, that is, the use of laser beams to generate a dispersion of NPs in a liquid environment, is emerging as a promising approach.^[4–7]

The oldest examples are dated back around 1991–1993,^[8,9]and are based on laser ablation in liquid (LAL),^[10] where a laser beam is directed on a solid target immersed in a liquid solu- tion, to generate a colloid through the ablation of the surface of the solid (Figure 1A). In most cases, LAL is performed with pulsed lasers, and it is also called pulsed-LAL (PLAL). The syn- thesis of nanomaterials by LAL is sometimes called laser abla- tion synthesis in solution (LASiS).^[11]In 1997,^[12]a variant of LAL appeared, where the laser beam is focused inside a liquid dis- persion of micrometric or nanometric powders, to obtain their photo-fragmentation into smaller NPs, in a process known as laser fragmentation in liquid (LFL, Figure 1B).[10,13, 14,243]

[a] Dr. V. Amendola

Department of Chemical SciencesUniversity of Padova Via Marzolo 1, 35131 Parova (Italy)

E-mail: vincenzo.amendola@unipd.it [b] Dr.-Ing. D. Amans

CNRS, Institut LumiHre MatiHre

Univ Lyon, Universit8 Claude Bernard Lyon 1 [c] Dr. Y. Ishikawa

Nanomaterials Research Institute

National Institute of Advanced Industrial Science and Technology (AIST) Tsukuba Central 5, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565 (Japan) [d] Dr. N. Koshizaki

Graduate School of Engineering Hokkaido University

Kita 13 Nishi 8, Kita-ku, Sapporo, Hokkaido 060-8628 (Japan) [e] Dr. S. ScirH, Dr. G. Compagnini

Department of Chemical Sciences University of Catania

Viale A. Doria 6, Catania 95125 (Italy) [f] Dr. S. Reichenberger, Dr.-Ing. S. Barcikowski

Technical Chemistry I and

Center for Nanointegration Duisburg-Essen (CENIDE) University Duisburg-Essen

Universit-tstr. 7, 45141 Essen (Germany) E-mail: stephan.barcikowski@uni-due.de

The ORCID identification number(s) for the author(s) of this article can be found under:

https://doi.org/10.1002/chem.202000686.

T 2020 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA.

This is an open access article under the terms of the Creative Commons At- tribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Part of a Special Issue on Low Temperature Solution Route Approaches to Oxide Functional Nanoscale Materials.

Soon, LFL turned out to be also an effective way for the re- duction of size and polydispersity of the colloids obtained by LAL. While performing LFL with laser pulse energy lower than the fragmentation threshold, it was observed that it is possible to obtain the photothermal fusion of aggregates of NPs into larger nanospheres,^[11,15]or the photothermal melting and va- porization of micrometric powders into submicron spheres,^[16]

through a process known as laser melting in liquid (LML) or pulsed-LML (PLML), Figure 1C.^[10,16]LML was also applied to the size increase of the colloids obtained by LAL. In fact, LAL, LFL, and LML can be combined for the production and the subse- quent size control (reduction or increase) of colloidal NPs, as shown already in 2007.^[17]Using milder fluence regimes to irra- diate a colloid with the intention mainly to change the atomic

Vincenzo Amendola is Associate Professor of Physical Chemistry at Padova University, where he obtained the PhD in materials sci- ence and engineering in 2008 and the Italian qualification as full professor in 2017, after re- search experience at M.I.T. and Cambridge University. With his Laser Assisted Synthesis and Plasmonics lab, he searches for uncon- ventional and non-equilibrium nanomaterials exploitable for experimental and theoretical investigations in plasmonics, sensing, nano- medicine and catalysis.

David Amans is Associate Professor of physics at University Claude Bernard Lyon 1. He stud- ied computing and materials science at Cen- trale Lyon from where he also received his Ph.D. in 2002. As a postdoc, he worked at University Libre de Bruxelles on quantum in- formation and nonlinear fibre optics, and then at PHELMA school on optronics. He joined the Institute of Light and Matter in 2005 where he is developing laser ablation in liquids, addressing nanomaterials science and laser-induced plasma.

Yoshie Ishikawa is a Senior Researcher at Nanomaterials Research Institute, National In- stitute of Advanced Industrial Science and Technology (AIST) since 2013. She obtained her Ph.D. from Kumamoto University in 2003 and was an Associate Professor in Kagawa University until 2013. Her scientific focus is on the fabrication and application of pulsed laser melting in liquid for metallic and inor- ganic sub-micrometre particle synthesis.

Naoto Koshizaki currently is Guest Professor in the Division of Quantum Science and Engi- neering of the Graduate School of Engineering in Hokkaido University. He obtained his Ph.D.

from University of Tokyo in 1997. Until 2013 he was a Senior Researcher in the National In- stitute of Advanced Industrial Science and Technology. Currently he works on physical fabrication methods for inorganic nanoparti- cles and nanocomposites.

Salvatore ScirH is an Associate Professor of In- dustrial Chemistry at the Department of Chemical Sciences of Catania University (Italy). His research activity is focused on het- erogeneous catalysis, with special interest in oxide-supported mono and bimetallic cata- lysts, and more recently to the application of photocatalysis to environmental protection and energy production. His activity is docu- mented by about 100 papers in international journals and books, 1 patent, and over 110 contributions in scientific meetings.

Giuseppe Compagnini is Full Professor of Physical Chemistry at the University of Cata- nia. His research group focuses on fundamen- tal and applied aspects of nanocomposites, including laser ablation, micro- and nano- joining, and vibrational spectroscopy. He is head of the Thin films and Nanostructures laboratory at the Department of Chemical Sciences and Head of the PhD School of Ma- terials Science and Nanotechnology. He is author of about 160 papers on international peer reviewed journals (ISI) and 10 invited re- views, receiving more than 5000 citations.

Sven Reichenberger is acting leader of the cat- alysis research group at the Institute of Tech- nical Chemistry of the University-Duisburg Essen. He received his Ph.D. in 2017 at the University of Duisburg-Essen and specialized in the field of laser-based defect engineering during a post-doctoral research. He currently focuses on his habilitation on surface process- es occurring during laser-based catalyst syn- thesis, with a focus on fuel cells, electrolyzers, and oxidation catalysts.

Stephan Barcikowski is Full Professor and Chair of the Institute of Technical Chemistry I at the University of Duisburg-Essen. In 2004 he received his Ph.D. in Mechanical Engineer- ing, followed by his habilitation in Chemistry on laser-generated nanomaterials in 2011. His research targets the nanoparticle formation mechanisms in laser ablation and fragmenta- tion, as well as their upscaling aiming at their application in catalysis, biomedicine, and ad- ditive manufacturing.

structure of NPs by introducing defects, while keeping the size unchanged, is called laser defect-engineering in liquid (LDL, Figure 1D). Inspired by the first seminal reports in the field, a growing international community of scientists pursued the study of physical and chemical processes involved in laser syn- thesis of colloids.^[6,10]They demonstrated several advantages of the method (Figure 1E) and that colloids can be generated with peculiar features (e.g., metastable phases and doped nano-crystals) not present in NPs obtained by other proce- dures, as will be described in the following text.^[6,10,18]

1) First of all, laser-generated colloids are highly pure, li- gands-free and expose an uncoated surface, because no chem- ical precursors, chelating agents or coordinating molecules are required in the majority of cases.^[11,19,20]Often, the absence of pollutant waste and the use of raw materials make laser syn- thesis compatible with the 12 principles of green chemistry,^[11]

offering new opportunities for the development of a green

and sustainable nanotechnology, and for the integration of col- loidal NPs in a circular economy.

2) The achievement of oxide NPs as a colloid does not expose the operator to airborne particle inhalation risks, thus allowing occupational safe and easy manipulation of the prod- ucts, compared to dry nano-powders.^[21] Besides, the effective interaction of NPs in a stable colloid with solid substrates and matrixes is facilitated by impregnation or mixing with the liquid phase.^[10,19]

3) Another relevant advantage of laser-assisted synthesis methods is the access to a wide range of oxide (and non- oxide) nanoparticles in similar experimental conditions, all at room temperature and pressure,^[18] as shown in Figure 1F on the basis of a literature overview. To clarify why laser synthesis falls under the “room-temperature and pressure synthesis” clas- sification, on the one hand, the mechanistically relevant local (temporal and spatial) temperature and pressure profiles and Figure 1. The landscape of laser synthesis of oxide nanoparticle colloids: Sketch of LAL (A), LFL (B), LML (C) and LDL (D). E) Honeycomb with frequently en- countered advantages in laser synthesis of colloids. F) Overview of oxide nanostructures generated by laser synthesis in liquid in the literature.

on the other hand, the practically relevant global-extrinsic pa- rameters have to be differentiated. It is worth specifying that locally and temporally, at the level of the matter interacting with the laser beam, extremely high temperatures and pres- sures are reached, but this is self-confined in a limited region of space coincident with the laser spot and the early explosive boiling volume. Molecular dynamic simulations coupled with the two-temperature-model were recently extended from the ultrashort-pulsed to nanosecond-pulsed laser ablation regime,^[22]where after short non-equilibrium phase the majori- ty of nanoparticles are in thermodynamic equilibrium with their local environment at the end of the simulation (a few nanoseconds).

These predictions are backed by experimental findings on binary, partly immiscible nanoparticle systems, where both a kinetic control and thermodynamic contribution of the particle formation dynamics are concluded.^[23–25] Hence, on the one hand, highly non-equilibrium conditions are pointing at a ki- netic control of the synthesis at the very early, sub-nanosecond formation time regime.^[26] Later, the whole system quickly reaches and remains in equilibrium with ambient conditions.

Globally, there is no need for strategies for heat or pressure regulation, which is a big advantage compared to gas-phase (pressure), hydrothermal (pressure, temperature), or sol-gel (temperature) synthesis, and wet chemical co-precipitation or reduction (temperature). Even with high repetition rate lasers (>kHz), liquid flux simultaneously works for draining the col-

loid and cooling the synthesis environment macroscopically keeping a steady state of temperature and pressure.

This means that laser synthesis does not require unit opera- tions for in-process-heating/cooling or pressure control, which makes laser synthesis systems easily implemented in laborato- ries. Overall, laser synthesis mechanistically benefits from ac- cessibility to metastable nanoparticle crystal structures or com- positions via temporal, pulsed, non-equilibrium condition that is confined in a microscopic volume, at the same time macro- scopically continuously operating in steady state, at room tem- perature and pressure.

4) By using the same equipment, it is possible to synthesize NPs in a few minutes and to switch from one type of nanopar- ticle to another with a „plug-and-play“ approach.^[21]

5) The experimental configuration can be tailored to the de- sired quantity of NPs, going from batch to flow cells (Figure 2).^[10] Flow cells have the advantage of limiting or avoiding the absorption of the laser beam by the just-formed NPs and persistent microbubbles,^[27] is especially relevant at visible and UV wavelengths.^[10,18]

6) The rich library of oxide nanoparticles obtained by laser synthesis in liquid also includes non-equilibrium phases and complex morphologies such as core–shell NPs, dendrites, spin- dles and, in specific cases, also nanowires, nanoflakes, nano- flowers, urchins, rods, sheets, and hollow spheres.^[18,35] As ex- plained in the following paragraphs, particle formation after each single laser pulse occurs on a timescale of 10^@6s,^[26]

Figure 2. Sketch of basic and advanced set-ups for the generation of colloids by LAL, LFL, or LML. Upper panel (from left to right, reprinted with permission from ref. [10], Copyright 2017, American Chemical Society): batch (ablation in beaker without stirring; reprinted with permission from ref. [28], Copyright 2015, American Chemical Society and ablation in beaker with stirring; reprinted with permission from ref. [29], Copyright 2015, Royal Society of Chemistry), semi-batch (reprinted with permission from ref. [30], Copyright 2014, American Chemical Society), flow (reproduced with permission from ref. [31], Copyright 2013, Royal Society of Chemistry) and flow-jet (Adapted with permission under CC BY 4.0 from ref. [32], Copyright 2016, Springer Nature Ltd.). The latter has been reported only for LFL and LML, see the following paragraphs. Lower panel: high throughput LAL set up with productivity up to >1 gh^@1(reprinted with permission from ref. [33], Copyright 2016, The Optical Society), remote-controlled flow LAL set up with embedded stop-flow optical control (reprinted with permission from ref. [34], Copyright 2019, AIP Publishing).

making possible the freezing of non-equilibrium phases or highly defective structures otherwise difficult to achieve.^[19,36–38]

7) A feature making laser synthesis very appealing for re- search and industrial exploitation is that the method is intrinsi- cally self-standing and can be implemented with minimal manual operation. For instance, it was recently demonstrated, that LASiS of various NPs is possible by controlling the equip- ment remotely with a PC or a smartphone.^[34]This is useful for an even more economic synthesis also for safer synthesis con- ditions as it minimizes operator permanence in nearby of the laser beam, flammable solvents or harmful compounds such as radioactive elements or volatile organic molecules, even when dealing with non-toxic oxide NPs.^[34]

Further developments in the automation of laser synthesis have been recently reported by real-time correlation of NP pro- ductivity with acoustic emission energy,^[39] as well as tempera- ture in the ablation chamber.^[40]

8) Laser synthesis in liquid is a linearly scalable method (pro- ductivity linearly scales with both laser power and time, at liquid flow operation), not yet demonstrated for production of kilograms of NPs or more.^[10,18]Nonetheless, the use of a high- power ultrafast laser with MHz repetition rate coupled with a polygon scanner (that achieves bypassing cavitation bubbles by supersonic lateral beam displacement) and a galvanometric mirror led to several grams/hour productivity of NPs by LAL, in a self-standing continuous flow set-up.^{[33, 41]}

9) Laser synthesis is generally considered an economically viable approach in the case of NPs involving precious metals or expensive compounds.^[42] Its economic viability is strongly dependent on material type. As a rule of thumb, the LAL pro- ductivity scales with the material density,^[43]making the lighter oxides less productive than, for example, the noble metals. Of course, also the ablation threshold fluence (often higher for oxides), the electron-phonon coupling as well as the (tempera- ture-dependent) target reflectivity contribute to laser ablation efficiency and thereby the power-specific productivity. But more material-specific factors contribute to the overall cost, in- cluding raw materials (bulk solids, solvents), hourly labour costs (manpower), the depreciation of the investment (equip- ment and its maintenance), the rental cost of the facilities, stor- age of products and the quality control of the whole proce- dure. The bottleneck for economic scale-up can only be identi- fied for a particular business case, but laser synthesis access to hundreds of different materials, making the detailed compari- son of economic viability for each of them out of the scope of this review. The commercial interest in laser-generated colloids is demonstrated by the existence of well-established compa- nies in Germany, Israel, and the U.S., commercializing this type of product for more than a decade.^[44–46]Besides, a few studies specifically addressed the case of laser-generated colloids and afforded some of the parameters in the previous list. Bench- marking between hydrothermal process, photochemistry and laser ablation in liquids has been conducted for low-priced oxide NPs (cerium oxide) dedicated to organophosphorus deg- radation.^[47] On the one hand, LAL was the most expensive method in this study on oxide nanoparticles, because of the usage of a not state-of-the-art laser setup with limited produc-

tivity of only 21 mgh^@1and the relatively high depreciation of the investment (equipment). On the other hand, LAL-generat- ed CeO2 NPs exhibited the best degradation activity because of the minor surface contamination inherent to the LAL-gener- ated NPs and were the only one amenable to in situ produc- tion without the need for high-temperature ovens. Conversely, for gold colloids, it has been calculated that the break-even point where laser synthesis beats chemical synthesis in the costs versus the mass of produced NPs plot already happens at tens of grams.^[42]This is possible for the limited manual op- eration, absence of expensive chemical precursors, and contin- uous advancement in synthesis scale-up. Interestingly, the lower cost of waste management was not even considered in the study. In both cases, the main source of cost in laser syn- thesis is connected to the laser equipment, that in the last 15 years showed a continuous growth of average power avail- able on the market, and a parallel decay of the equipment cost per watt,^[10] suggesting a favourable prospect for further reduction of production costs in case of laser-assisted synthesis of colloids.

In this review article, we provide the working principles behind LAL, LFL, LML, and LDL focusing on oxide nanostruc- tures, with an overview of nanomaterials produced so far by this method, and of their functional properties and reported applications. So far key reviews have mainly focused on metals, alloys or processing variants during LSPC, while oxides were mentioned only peripherally.[5,10,19,36,48,49]By definition, an oxide nanomaterial consists of a nanoscale object including oxygen atoms in its chemical formula. Generally, in this review, oxide materials are defined as inorganic compounds obtained by reaction of oxygen with an element with low reduction po- tential. The redox potential changes with environmental pa- rameters such as solvent, solute (concentration), temperature and pressure and, in fact, also noble metals can be surface-oxi- dized in some laser synthesis conditions.^[19] However, this is usually limited to a minority of the atoms in the NPs, therefore the chemical composition of the resulting material only con- tains a relatively small amount of oxygen. Hence, laser-generat- ed and laser-processed nanomaterials that only express limited oxidation are excluded by this review, which deals only with the compounds where the content of oxygen is comparable to that of the other main elements.

It is expected that the research on laser synthesis for the preparation of oxide nanoparticles in liquids will continue to grow in the near future, especially if one considers the variety in terms of composition and structures that are achievable, the interest in better control of the composition, size, morphology, and phase, and the need for improving the understanding of NP formation mechanism. To support this development, a non- profit conference series on Advanced Nanoparticle Generation and Excitation by Lasers in liquids (ANGEL)^[50]exists since more than a decade, and handbooks about laser synthesis and proc- essing of colloids are available through open access for begin- ners,^[21]in addition to multiple specific and advanced review ar- ticles that appeared in recent years.[10,11, 16,18–20,26,35, 3648,52–54]

In the following, the review will introduce the fundamental concepts of LAL and give an overview and discussion of the

work and perspectives of laser-based synthesis of conventional nano-oxides, core–shell NPs, defect-engineered oxide NPs, and ligand-stabilized oxide NPs. The discussion will include also the synthesis of multicomponent oxide nanostructures by sequen- tial LAL and reactive LAL, highlighting the issues encountered with the compositional homogeneity of the products. Subse- quently, the fundamentals and an overview of oxide NPs ob- tained by LFL and LML as well as upscaling considerations will be treated. Progress in the recently established laser-based defect engineering in liquid (LDL) will be also discussed.

Similar to LFL and LML, the LDL method treats dispersed particles, with comparable mild excitation conditions, but not primarily intending to downsize particles (like LFL) or total par- ticle melting (like LML). Instead, LDL aims at defect introduc- tion, also relevant for the preparation of (surface-)doped oxide NPs. The review is concluded by an overview of recent, most relevant applications of laser-generated oxide nanomaterials, with special emphasis on photo-catalysis, oxidation-catalysis, bio-applications, and photonics.

LAL

LAL fundamentals

Before entering a more insightful description of the ablation mechanisms, it is useful to consider first the major steps of the most prominent laser synthesis method, the LAL (Figure 3).

Starting from the general LAL configuration where a laser beam is focused on a bulk target immersed in a liquid, first of all, the laser beam should travel through a low-absorbing liquid layer, which means that the liquid must be transparent at the chosen laser wavelength, and liquid breakdown must be avoided at the fluence selected for the experiment.^[10,18] This issue is common to LAL, LFL and LML as well, and it becomes especially challenging when using ultrashort pulses, due to the self-focusing and filamentation effects.^[55–59]

Assuming the requisite of liquid transparency, the interaction of the laser beam with the bulk target results in the formation of a plasma in a few hundreds of picoseconds.^[60–65]The plasma is initially made of the target material. Because of the fast volume expansion (appearing already after about 10–

100 ps),^{[57, 66]} a mechanical shockwave is released both in the target and in the liquid (Figure 3).^[67–71] Such a shockwave can lead to phase transition at the target and the pressure at the focal point can reach a few gigapascals.^[60] On the other hand, the interaction of the plasma with the liquid leads to the fast vaporization of the liquid, which is observed already at the shortest gating times (few ns) of CCD cameras commonly used to observe plasma dynamics. A cavitation bubble is initiated and appears mainly composed of the vaporized solvent.^[61,71–73]

As depicted in Figure 3, the cavitation bubble grows and col- lapses after a characteristic time depending on the pulse energy and the pulse duration (typically a hundred microsec- onds for nanosecond pulses of a few mJ). When the cavitation bubble collapses, the NPs are released in the liquid.^[74]Howev- er, crystalline particles have been observed also outside the ex- panding bubble, preceding the bubble’s expansion front, by small-angle X-ray scattering (SAXS) experiments,^[75] confirming the theoretical predictions made by Zhigilei et al.^[76]Due to the high amount of energy accumulated in the point of bubble collapse, rebound cavitation bubbles can grow and collapse again, depending on system parameters such as liquid me- chanical properties and initial bubble energy. The Rayleigh–

Plesset equation is only suitable to describe the first oscillation, while the Gilmore model including the liquid compressibility is required for modelling the subsequent oscillations (re- bounds).^[77] Note that even Gilmore model cannot adequately describe bubble dynamics with broken symmetry. In detail, the LAL bubble neither has as a perfect hemisphere aspect ratio (which dynamically strongly deviates from 0.5, in particular at expansion and collapse phase) nor has a circular contour. The bubble contour’s circularity is broken at the interface layer di- rectly on top of the target, so a LAL bubble geometry can—

Figure 3. The timeline in LAL for ultrashort pulses shows the successive steps that occur during LAL synthesis, from the laser pulse interaction with the target to the release of the as-produced NPs in the solution. On top, the characterization techniques are displayed according to their temporal resolution. At the bottom, from left to right: electron-phonon coupling scheme, phase transition snapshot from molecular dynamics simulation, optically active plasma and re- leased shockwave, bubble dynamics and produced colloidal solution. Reprinted with permission from ref. [60], Copyright 2017, Elsevier.

simplified—be divided into the root part on which a quasi- hemispherical cap sits. These effects become quite obvious at smaller bubble sizes (i.e., smaller LAL pulse energies) and are particularly expressed in high viscosity liquids.^[78] After bubble collapse and NP release in the liquid, NPs may further oxi- dize^[79] and slowly grow on a relatively limited extent or may undergo agglomeration if they have limited colloidal stabili- ty.^[10,18]Importantly, the requisite of transparency at laser wave- length holds also for the laser-generated NPs, that may reduce productivity per pulse due to absorption and scattering (and by strongly reducing the Kerr limit for optical breakdown). In fact, the ablation rate is higher when using low wavelengths as long as no NPs synthesized by previous pulses are present (Figure 4 A–C).^[10,80]Besides, it was demonstrated that produc- tivity is affected also by persistent microbubbles stemming from vaporization or even degradation of the vapour layer at the boundary with the plasma plume,^[27]so that huge amount of permanent gases are formed proportional to the redox po- tential of the target.^[81] It is worth noting that laser beam ab- sorption by NPs (known as beam self-absorption) may induce structural and chemical changes to the NPs (LFL or LML like).

Structural and chemical changes are associated with broaden- ing and polymodality of the size distribution, as well as to phase heterogeneity of products.^[10,18]To avoid or limit the self- absorption of the primary beam, infrared or near-infrared wavelengths, and LAL with a continuous flow set up (see Figure 2), are usually preferred.

Besides, when the productivity needs to be pushed on the gram scale, sufficiently high laser fluence (Figure 4D–F), low pulse duration (Figure 4G–I) and high repetition rate lasers (>

kHz, that is, inter-pulse delay of < 1 ms) have shown to be most successful up until now.^[10,80]

But, it is worth to stress that limitation to the general trends reported in Figure 4 needs to be considered. For instance, the dependence of productivity versus fluence (Figure 4F) depicts

the regime where the fluence is below &7.4 times (e²) the ablation threshold found (fth) to be valid for fs-, ps- and ns- laser pulses.[33, 82,83]When the fluence reaches fth·e²the produc- tivity was found to reach a maximum.^[33,82] Here, the former was initially predicted by the theoretical model of Neuensch- wander et al. linking the increase of ablation rate with the in- creasing optical penetration depth into the target (with rising laser fluence).^[84]At laser fluence above fth·e²the optical pene- tration depth is limited^[84] and hence (further increasing) only leads to a stagnating or even decreasing productivity.^[33,82] In this context, self-focusing and filamentation effects in the liquid medium, occurring especially for high laser intensities (>10¹³Wcm^@2,^[56] usually only reached with ultra-short pulses such as fs up to several ps), may further decrease the produc- tivity at high laser fluence as significantly less energy reaches the target in this case.^[82]

In general, the laser fluence can be increased by varying the pulse energy of the laser (Figure 4D) or decreasing the beam spot size by optimizing the working distance between the target and focusing lens (Figure 4E). Again note that, if the spot size becomes too small, the productivity will drop,^[10,80]

since the penetration depth of light into the matter is limited by the material’s absorption properties (temperature-depen- dent reflectivity). In other words, a linear increase in fluence is not compensated by a linear increase in the ablation efficien- cy.^[10,85]Consequently, increasing the laser fluence by lowering the spot size from the mm scale to the micron scale will lead to a smaller ablation rate (Figure 4F).^[10,80] Additionally, espe- cially from the work of Kautek and co-workers,^[86]the beneficial effect of material defects in lowering the ablation threshold is known for decades. With a smaller spot size, generally less de- fects would interact with the laser pulse such that a higher ablation threshold (and thereby a lower productivity) can be expected with decreasing spot size.^[86]

Figure 4. Illustrations on the change of nanoparticle productivity with the applied laser parameters; (A–C) laser wavelength; (D–F) incident laser fluence; (G–

I) laser pulse duration (single-pulse productivity) and (J–L) inter-pulse distance. The left pictures (A, D, G, J) illustrate the effect of the respective parameter in the absence of NPs; the middle (B, E, H, K) in the presence of NPs during laser ablation. Illustrations to the right (C, F, I, L) show generalized trends occurring to productivity when varying the respective laser parameter while keeping the other parameters constant. (adapted with permission from ref. [80]).

In case of pulse duration, two processes need to be consid- ered for productivity: 1) thermal loss, which is highest in case of longer pulse duration (>10 ps, see Figure 4G) and 2) plasma shielding (Figure 4H) which is especially pronounced for ns laser pulse duration and above (in line with plasma dy- namics, compare Figure 4I and Figure 2).^[10,80] Hence, highest ablation efficiency is usually predicted in case of fs- and ps- laser ablation as depicted in Figure 4I.^[10,80] While cavitation bubble shielding was neglected by employing sufficiently opti- mized scanning speed in both cases, the productivity, there- fore, seems not only to decrease with increasing pulse dura- tion but it appears that there exists an additional sweet spot of pulse duration, material and pulse energy.

Hence, at the same nominal fluence (the effective fluence decreases with increased pulse duration) there will be a materi- al-specific optimal laser pulse duration that also avoids losses by NP absorption, for example, 2 ps for LAL of gold.^[87] Above the kHz frequency threshold, the lifetime of the cavitation bubble exceeds the inter-pulse delay (Figure 4J–L), so that the laser beam propagates through a highly scattering liquid/gas interface that is responsible for a sensible reduction in the ablation efficiency. When the inter-pulse distance is too short, the subsequent laser pulse and the cavitation bubble generat- ed from the previous pulse overlap in time (Figure 4J). In this case, extensive scattering of the laser pulse occurs and, hence, the productivity decreases (Figure 4L).^[10,80]These issues were solved by Barcikowski’s group using a polygon scanner (com- pare Figure 2) coupled with a single galvanometric mirror, al- lowing the laser beam to spatially bypass the cavitation bubble (Figure 4K) at MHz repetition rate and laser power

&500 W (ps-laser). The polygon scanner allows supersonic scan-rates and ablates a different position of the target with each pulse.^[33,41]Ablation rates of several grams per hour were reached with this high-end class laser type.^[83,88] Interestingly, Dittrich et al. have shown that the ablation efficiency (but not the absolute productivity) of ultra-low-power and comparably cheap compact ns-lasers is a factor of 8 higher compared to the high-end class ps-laser.^[83]

Mechanistic insights

The processes involved in laser ablation of solids in liquid envi- ronment have been the subject of extensive studies,^[86]with a special focus on non-thermal processes and the early stages,^[26,89]usually with metallic targets and ultrashort pulses.

Many efforts were made to develop in situ characterization meth- ods and numerical simulations, leading to the conclusion that the processes occurring at the early time scales after laser energy deposition are critical in the definition of the final product.^[26]

Figure 3 also lists the different characterization methods re- ported in the literature with the reachable time scale for each of them. These measurements gave evidence on the early gen- eration of the NPs. Light scattering experiments,^[90,91]as well as in situ time-resolved small-angle X-ray scattering (SAXS),^[75,92]

not only showed that the NPs are confined inside the vapor bubble but also showed that NPs are present earlier than bubble formation. In particular, light scattering experiments

suggest that NPs are present after a few hundreds of nanosec- onds.^{[90, 91]}This is consistent with the fast cooling of the plasma reported from plasma spectroscopy (10 Kns^@1)^[60] or depicted in modelling.^[22,93,94]

The fast cooling of a laser-generated hot gas or plasma com- monly leads to nucleation and growth of particles.^[53,95,96] The standard pathway to the formation of nanoparticles generally includes three stages (Figure 5A): nucleation, evolution of nuclei into seeds, and seed growth into final nanocrystals. Al- though a general picture of how these steps evolve in LAL is still under construction, Zhang and Liang argued that the time required to reach the critical concentration for nucleation in LAL is much shorter than that of the wet-chemistry synthetic routes, due to the fast ejection dynamics of the „precursors“,^[53]

and related large temperature gradients of up to 10¹²Ks^@1. Moreover, as a large part of the process takes place in the gas phase (cavitation bubble), or at relatively low concentration once in the liquid phase, where particles have slow mobility, particle growth by coalescence and ripening may last for a longer time than in the conventional La Mer mechanism used for wet-chemistry methods.^[53] Some relevant open points for oxide nanoparticle formation concern the chemistry that is dis- cussed in the next paragraphs. This echoes the questions of the physicochemical interaction which must be addressed to understand the fast vaporization of the solvent, as well as the parameters favouring its decomposition and its reactivity.

The scenario of Figure 5A is supported by molecular dynam- ics simulations developed by Zhigilei’s group.[22,76,93,94] They have developed atomistic simulations of the laser ablation of metal targets in water, combining a coarse-grained representa- tion of the liquid environment and an atomistic description of the laser interaction with metal targets and for pulse durations from fs up to few ns.[22,76,93,94] For ultrashort laser pulse dura- tions a bimodal size distribution is predicted, as frequently re- ported from transmission electron microscopy,^[97]and also from SAXS measurements.^[74,75,92]

In the case of hundreds of ps to ns pulses, the thermal and stress confinement characteristic of ultrashort pulse laser abla- tion is not observed.^[22] This is associated with different abla- tion dynamics and partly explains why the nanoparticle size distribution tends to be broad in ns LAL but not bimodal as in ultrashort-pulsed LAL, as described in Figure 5B.^[22] At the same effective fluence, independent of the pulse duration (ps or ns), three NP formation pathways are predicted by the works of the Zhigilei group, which are linked to the regions where they originate from, summarized as follows along three regional sections, from the target towards the liquid: 1) The nanoparticles mostly emerge from the spinodal decomposition of a part of the ablated material located between the target surface and the transient interfacial metal layer. High density of ablated matter (and low amount of supercritical water) is characteristic for this region. High temperature causes the seeds (1-4 nm) to be thermodynamically unstable and there- fore evaporate unless their rapid collision and coalescence leads to small (about 5 nm) nanoparticles, while the larger nanoparticles in this region continues to grow. These processes result in thermodynamically stable, mostly larger nanoparticles

of around 5–10 nm, although some smaller ones survived the growth process. 2) The decomposition of the thin transient metal layer at the interface between the ablation plume and water environment causes the generation of large (>10 nm) molten nanoparticles, only slowly cooled by surrounding su- percritical water. 3) At the very front of the emerging cavita- tion bubble, evaporation from the hot interfacial layer into the supercritical water causes the formation of very small (<5 nm) nanoparticles through the nucleation and growth from the vapor-phase metal atoms. These small nanoparticles solidify in nanoseconds, likely in defect-rich nanocrystals, whereas parti- cles stemming from pathways 1) and 2) are still in molten stage after a few nanoseconds. In Figure 5B), one obvious dif- ference that shorter pulse duration causes is jetting of particles stemming from the pathway 2) as a result of a more vigorous ablation process, which includes the ejection of metal droplets directly into the high-density water region. Note that the intri- guing sketches in Figure 5B) are intended to explain the mech- anism^[22] but they are not to scale at all. In reality, after some ns the height of this early formation volume is only hundreds of nm, whereas the width is the laser spot size (tens of mi- crons), so the virtual picture one should have in mind is a very flat object with an aspect ratio of about 100.

Noteworthy, there are several additional complexities when considering oxide targets instead of metal ones. First, there is

a lack of optimized empirical potentials for metal oxides which could be effectively used in a molecular dynamics simula- tion.^[98]The reasons are the increased complexity of the intera- tomic potentials when several chemical elements are involved (at least the oxygen and a metallic element), and the poly- morphism issue since the metal oxides usually form various stoichiometries and crystal structures. Second, while in metals there is a direct laser-heating of the free electrons in the con- duction band, in metal oxides electrons must be promoted across the band gap by the laser excitation before their heat- ing (as discussed below). The modelling of the laser-target in- teraction is then more challenging than for metals. Third, mo- lecular dynamics are not suited yet to catch the chemistry in the early ablation phase leading to metal oxide NPs. Fourth, numerical simulation assuming ultrashort pulse duration cannot catch the whole complexity of the processes resulting from the plasma-laser interaction which occurs for (several) nanosecond pulse duration, although recently a convincing modelling approach in that direction has been presented.^[22]

Indeed, for nanosecond pulses, shielding of the target by the plasma occurs, with an increased effect with the decreas- ing wavelength.^[99]In addition, the surface structure such as its roughness changes significantly also on the ns time scale, which is expected to locally modify the optical properties of the target and, thus, its light absorption properties.^[22] There- Figure 5. (A) Sketch of NPs formation through ablation (laser-induced jetting) of the target material (that in general may happen as fragments and large drop- lets), growth and coalescence, ripening and, eventually, agglomeration if the system is not colloidally stable. (B) Sketch of the processes initiated by irradiation of a metal target in water by an ultrashort (top, femtosecond to tens of picosecond) or short (bottom, hundreds of picoseconds to nanoseconds) laser pulses.

Blue: liquid; grey: metal target; light grey: ablation plume; light blue: cavitation bubble precursor; dark blue: ejected materials. For description see text. Re- published with permission from ref. [22], Copyright 2020, Royal Society of Chemistry. (C) Timescales of various electron and lattice processes in laser-excited solids.^[89]Each bar represents an approximate range of characteristic times consistent for carrier densities from 10¹⁷to 10²²cm^@3. (D) Sketch of main differen- ces in laser ablation of metallic and oxide targets.

fore, the energy deposited on the target, as well as plasma warming are difficult to account quantitatively.

Despite the apparent complexity of the processes, some evi- dence comes from the experimental investigation of the abla- tion processes, including ablation of semiconductors and di- electrics. Figure 5C shows the timescales of various electron and lattice processes in laser-excited solids for an ultrashort pulse (femtosecond). The subsequent typical pathways of energy dissipation and phase transformations following the ex- citation by an ultrashort pulse are also displayed in Figure 5C.

As anticipated above, the main difference between metals and metal oxides lies in the mechanisms of electron excitation due to the large energy band gap of the latter, leading to several differences in linear and nonlinear absorption (especially at NIR wavelength), as well as in electron dynamics. Overall, the abla- tion of metal targets could appear more convenient, which jus- tifies why most works dealing with the synthesis of metal oxides are based on the oxidative laser ablation of bulk metals (see next paragraph). Another reason for this choice is related to the mechanical properties of the materials, since pure metals are commonly more ductile than their oxides. Brittle metal oxides are more subject to shockwave-induced damage, which leads to target-crushing and target break-up (Fig- ure 5D). As a consequence, the mass of NPs produced by laser ablation of bulk oxide targets can strongly differ from the mass removal from the target, requiring additional purification steps to remove the unwanted target pieces. For instance, the ablation of a Gd2O3target with an Nd:YAG laser source (500 ps, 2 mJ/pulse, 1064 nm, 2V10¹¹Wcm^@2) led to the production of NPs (diameter < 100 nm) of 2.00 :0.18 ng/pulse, which howev- er corresponds to only 13% of the removed material from the target (&15 ng/pulse).^[100]In this context, the porosity of press- ed YIG (yttrium iron garnet) micro-powder targets was shown to be detrimental, since a high porosity leads to a large frac- tion of microparticles detaching from the target and being present in the colloid.^[101] Conversely, the ablation of a high- density powder target (>99 %) led to the formation of &3 nm YIG NPs similar to the ablation of a bulk YIG target.^[101]

Concerning the ablation mechanism of oxide targets with common laser sources used in LAL, such as the 1064 nm nano- second pulses of Nd:YAG laser in fundamental mode, the photon energy (1.17 eV) is lower than the band gap energy of most oxides. As an example, the energy of 8 photons is needed to cross the Al2O3 band gap with a 1064 nm laser source. The promotion of the electrons from the valence band thus involves photoionization processes preceding avalanche ionization. Photoionization acts as an initial step in the laser energy deposition and subsequent material modifications, leading to the optical breakdown of the solid.^[102] There are two different processes of photoionization, multiphoton ioniza- tion and tunnelling ionization.^[103]The multiphoton excitation involves the simultaneous absorption of several photons. The ionization by electron tunnelling is induced by a distortion of the potential barrier for large laser fields, and it is relevant for high electric field intensities, as those reached with ultrashort laser pulses.

A better understanding of the ablation conditions is provid- ed by the value of the parameter g, which gives the balance between the multiphoton ionization regime and the tunnelling ionization regime and is defined as [Eq. (1)]:

g ¼2pc l

ffiffiffiffiffiffiffiffi pmD

eE ð1Þ

where l is the laser wavelength, c light speed, m the re- duced mass of the electron, e the electron charge, D the energy band gap of the material, and E the magnitude of the laser electric field, which scales with ffiffiffiffiffiffiffiffi

p1=t

, where t is the pulse duration.^[104] When is decreased below 1, the optical breakdown is governed by the tunnelling ionization regime,^[105]

that is regarded as a „deterministic breakdown regime“, that is, with high precision in the spatial delivery of pulse energy to the target.^[103,106] According to Equation (1), decreasing g to reach a deterministic breakdown regime can be achieved by increasing the wavelength and decreasing the pulse duration.

Once electrons are promoted to the conduction band, they can absorb laser energy through inverse bremsstrahlung (elec- tron acceleration). For ns pulses at ordinary fluencies, electron acceleration is equilibrated by electron-phonon scattering.

Conversely, for ultrashort pulses, electrons are accelerated up to the threshold for achieving avalanche multiplication by elec- tron-electron scattering, thus further increasing the electron density in the conduction band.

For semiconductors, one can imagine that the initial pres- ence of free carriers (doping) could help to decrease the abla- tion threshold. For dielectrics, optical defects could also help to decrease the ablation threshold. However, Leyder et al. have compared the non-linear absorption inside silicon for samples with different initial free-carrier densities,^[107]that is, for doping concentrations from 10¹³cm^@3to 10¹⁸cm^@3. For a 130 fs pulse at 1.3 mm, their results demonstrate that the laser energy dep- osition does not depend on the doping concentration, and thus the avalanche is not efficiently triggered even up to a 10¹⁸cm^@3 free electron density. The multiphoton excitation is a nonlinear process highly sensitive to the laser intensity I, lead- ing to a scaling law I^Nfor the absorption with N the number of photons involved.^[108]The ablation threshold is decreased at a shorter wavelength, with a consequent increase in the ablation rate.^[108–110]

To increase LAL efficiency with oxide targets, it could appear convenient to use short wavelengths in the near UV, but with the drawback of self-absorption of the laser beam by the pro- duced NPs (see Figure 3B–D). Another approach to improve the ablation efficiency relies on promoting material breakdown in the tunnelling ionization (deterministic) regime instead of the multiphoton one.

On the other hand, the decrease in pulse duration is associ- ated with the issue of self-focusing and liquid breakdown or filamentation for such high fluences (for instance the threshold of optical breakdown in water is 1.11V10¹³Wcm^@2).^[111]Recent- ly, DoÇate-Buend&a et al. have elegantly overcome the issue of the filamentation and non-linear energy losses in the water when femtosecond laser sources are used in LAL.^[55]They have

applied the simultaneous spatial and temporal focusing (SSTF) of femtosecond pulses configuration, that avoids the unwant- ed non-linear effects, and ensures a thigh control of the abla- tion spot for a femtosecond laser source (45 fs pulse duration, 800 nm, 1 kHz, 200 mJ/pulse, 7V10¹³Wcm^@2at the focal point in water).^[55]Though promising, the best productivities to date remain those obtained with high repetition rate ns and ps laser sources coupled with scanning optics to bypass the cavi- tation bubble.^[33,112]

Overview of oxide NPs obtained by LAL Conventional nano-oxides: role of bulk target.

The working principle of LAL seems to suggest that the easiest way to produce oxide NPs is starting from a bulk oxide target.

However, this is the less frequent case found in the literature, where the majority of works report the laser ablation synthesis in liquid solution of oxide NPs starting from a bulk target of pure metal. In fact, it is worth noting that in LAL, the matter extracted from the solid target ineluctably encounters the mol- ecules of the liquid solution, in three different conditions which are (Figure 6A): 1) interface between the ablation plume and the surrounding liquid, 2) the interior of the cavitation bubble, in the gas phase and 3) the liquid at ambient tempera- ture and pressure after the collapse of the cavitation bubble.

The formation of radical species during laser-induced break- down of solvent molecules in the plasma at target surface has been intensively investigated in the last years,[27,113–115]showing that these radicals may react forming persistent microbubbles consisting amongst others of H2, O2,^[27,115] and H2O2[114] when LAL is performed in water. Excited oxygen species have been observed in real-time inside the plasma plume in aqueous en- vironment, up to hundreds of nanoseconds after pulse absorp- tion,^[116] and also in ambient air during ablation of oxide tar- gets.^[117] Hence, oxygen coming from the molecules of liquid (e.g., H2O) or additives (e.g., H2O2or atmospheric O2) will react

with the ablated target species, and the extent of the oxida- tion reaction will depend on the type and concentration of re- active oxygen species and on the redox potential of the metal.^[81]This is the source of persistent microbubbles affecting the ablation rate.^[27] For the ablation of 7 different metals, Kalus et al. observed that the developed gas volume is directly correlated with the respective redox potential of the metal.^[81]

A possible correlation of (temperature-dependent) redox po- tential and oxidation state during LAL has been discussed re- cently in literature.^[19]

All this makes the choice of the bulk target crucial for the achievement of the desired oxide NPs. The case of iron nano- oxides is useful to exemplify this aspect, given the variety of possible iron compounds and the relatively simple characteri- zation.^[118,119]It has been reported that ns-laser ablation of bulk metal Fe target in water gives a prevalence of magnetite Fe3O4

NPs, with a minority of hematite a-Fe2O3, wustite FeO and even some traces of metal Fe, likely present as a core inside a protecting oxide shell.^[118] However, when ns-laser ablation is performed with a hematite target in water, amorphous Fe2O3

particles are collected.^[120] Conversely, maghemite (g-Fe2O3) nano-oxides are obtained from the hematite target in ethanol or acetone, which are known to have a partially reducing effect on the ablated material during LAL.^[69,121]This is in agree- ment with computer simulations and experiments performed on alumina, indicating that a slight excess of oxygen is re- quired to achieve oxide NPs with the same stoichiometry of the target when ablated material is still in the gas phase of the cavitation bubble.^[117]

This suggests that LAL of oxide targets in organic solvents such as alcohols or acetone give slightly oxidized NPs, while oxidation is promoted in water eventually forming amorphous and hydroxylated compounds. Nonetheless, crystalline oxide NPs can be obtained by laser ablation of crystalline oxide tar- gets in water, as demonstrated with 1064 nm (13 ns) pulses and a 98 wt.% ZnO: 2 wt.% Al2O3 target in MilliQ water.^[122]

Figure 6. A) Sketch of LAL highlighting the three environments where the target species encounter oxygen species: in the plasma plume (1), in the cavitation bubble (2) and in the liquid at ambient conditions (3). B) By changing the LAL parameters listed in (A), it is possible to switch from AgO microcubes to metal Ag (adapted with permission from ref. [123], Copyright 2011, American Chemical Society, from Fe3O4(adapted with permission from ref. [119], Copyright 2011, Royal Society of Chemistry), to Fe-Fe3O4(reprinted with permission from ref. [157], Copyright 2011, American Chemical Society), from CuO/Cu2O/Cu (re- printed with permission from ref. [180]), to Cu-CuO (reprinted with permission from ref. [173], Copyright 2019, Elsevier), or from ZnO (adapted with permis- sion from ref. [144], Copyright 2005, American Chemical Society) to Zn-ZnO NPs (adapted with permission from ref. [142], Copyright 2005, American Chemical Society).

Conversely, LAL of metal targets in aqueous environments gives oxides, sub-stoichiometric-oxides or hybrid metal-oxide structures. These cases are discussed in better detail and with the help of specific examples in the next paragraphs.

Conventional nano-oxides: role of oxide type and LAL parameters

In several cases, LAL allows tuning the composition and the structure of oxide NPs, ranging from compounds where the metal has the largest possible oxidation state, to core–shell structures where only an external shell of oxide is formed around a core of pure metal (Figure 6B). The presence of oxygen atoms in the plasma plume is associated to highly oxi- dative conditions, such that even noble metal (Ag,^[123]Au,^[124,125]

Pt,^[126]Pd,^[127] Rh^[128,129]) NPs with a fraction of oxidised atoms have been observed. Oxidation is promoted by the use of ns laser pulses, UV wavelength and ion stabilizing additives, com- pared to pulses with a shorter duration, use of NIR wavelength, or presence of additives not interacting with metal ions. For in- stance, Ag2O nanocubes were obtained with 248 nm (30 ns) LAL of bulk Ag in aqueous solutions of polysorbate, while metal Ag NPs with limited surface oxidation are obtained in similar conditions without adding polysorbate,^[123] or at 1064 and 532 nm in aqueous solutions of sodium dodecyl sul- phate.^[130]UV light is re-absorbed by the NPs, giving a simulta- neous process of LAL and LFL, with the result of increasing the chance of reaction between metal and oxygen atoms. The role of the ion stabilizing additive is that of slowing down the rate of electron transfer from reducing species to coordinated metal cations in solution, thus increasing the probability of re- action with oxygen.^[123]

Chemical oxidants can be added to the solution to promote the reaction of target atoms with oxygen, as shown by LAL of bulk Cu in pure water and aqueous solutions of H2O2

(1–5 vol.%) with 532 nm (5 ns) laser pulses, obtaining, respec- tively Cu2O or CuO nanocrystals.^[131]LAL with 355 nm ns pulses in 3 vol.% H2O2has been reported also with a Ni target to ach- ieve NiO NPs.^[132] In another report, gallium oxide Ga2O3 has been found after LAL of pure GaAs in acetone with 532 nm (7 ns), while non-oxidised GaAs NPs were achieved when using 250 fs pulses in the same conditions.^[133] LAL with 1064 nm (10 ns) pulses of a GaN target in water also originated GaNO NPs.^[134]Regarding the pulse duration, the different results may be ascribed to the longer lifetime of the plasma plume when longer laser pulses are used. This is associated with a more ex- tended mixing of target and solution species in the highly re- active plasma conditions,^[10,18,26]as well as in a longer lifetime of the cavitation bubble and potentially prolonged persistence of NPs in the gaseous phase at temperature @ than room tem- perature.[26,77,135,136]In fact, there are many reports where LAL with ns pulses of metal targets in water produced oxide NPs.

For instance, g-Al2O3nanocrystals co-doped with H⁺ and Al²⁺

,^{[137, 138]} Co3O4,^[139] Fe3O4,^[118,119] TiO2[29] and MoO3[140] NPs were produced by 1064 nm ns pulses starting from a target of, re- spectively, metal Al, Co, Fe or Mo in water. Wurtzite ZnO NPs were produced with 1064 or 532 nm ns pulses starting from a

Zn target in water.^[141–144] Photoluminescence measurements showed that ZnO particles obtained by LAL can be rich with oxygen vacancies.^[145]ZnO particles co-doped with Al2O3have been synthesized starting from a bulk ZnO target doped with 2 wt.% of alumina, dipped in water, and using 1064 nm ns pulses.^[122] Recently, non-oxidized Zn atoms from a Zn target ablated in water with 7 ns–1064 nm pulses were detected inside the cavitation bubble still after tens of microseconds by in situ x-ray absorption spectroscopy, and the metal signature prevailed even for milliseconds (i.e. after bubble collapse).^[51]

SnO2was produced by 355 nm (10 ns) LAL of Sn in water.^[146]

Analogous results were reported when using longer laser pulses of 6 ms at 1064 nm and a target of Gd in diethylene glycol, which produced Gd2O3 NPs by the reaction of ablated Gd atoms with atmospheric oxygen dissolved in the liquid and oxygen coming from solvent pyrolysis.^[147] The use of laser pulses with longer duration is likely to further extend the plasma lifetime and volume compared to ns-pulses, facilitating the ionization of solvent molecules and the reaction of target and solution species in the plasma or nearly plasma condi- tions.[26, 77,135] Interestingly, the average size of Gd2O3 NPs is 4 nm, significantly lower than the average size of oxide NPs obtained by LAL in water (at the same pulse duration of 6 ms), that ranged between 10 and 30 nm.[118,119,137–139]This is attribut- ed to ethylene glycol properties, such as adsorption on the surface of Gd oxide clusters and high viscosity hindering clus- ter coalescence.^[18,147] Oxidation of target species occurs also during LAL of organic materials like coal in ethanol and 355 nm (10 ns) pulses, giving graphene oxide quantum dots.^[148]

When LAL is applied to oxide target, the surrounding liquid may influence the crystallinity of final products. This has been shown especially for the LASiS of rare-earth-doped oxides, such as YVO4:Eu³⁺NPs obtained with 532 nm (10 ns) pulses in water, ethanol or mixtures of the two liquids.^[149]The different interaction of solvent molecules with the surface of YVO4:Eu³⁺

NPs was evident from the achievement of crystalline ovoidal particles in water versus spherical and partially amorphous par- ticles in presence of ethanol,^[149]or spherical crystalline parti- cles in an aqueous solution of SDS.^[150] This suggested that YVO4:Eu³⁺nanocrystals are stabilized more effectively by etha- nol molecules than by aqueous surfactants like SDS. Indeed, the reactivity of organic solvents with the ablated species needs to be considered for each specific material and set of synthetic parameters. For instance, LASiS of GaO colloids has been reported with 1064 nm (10 ns) pulses and a GaO target in ethanol,^[151] or defective CeO2 nanocrystals were produced by 1064 nm (10 ns) pulses and a CeO2 target in water.^[152] In- stead, in the case of Ti target and 1064 or 532 nmns pulses, the oxide phases or rutile and anatase TiO2were achieved only in water, while TiC was found in alcohols.^[153–156] Also in the case of Fe targets and 1064 nm (10 ns) pulses, magnetite (Fe3O4) was found in solvents like acetonitrile and dimethylfor- mamide, while a mixture of magnetite and iron carbide (Fe3C) was obtained in ethanol.^[157]Analogous results were found in LAL of Fe with fs pulses.^{[158, 159]}